AIR

AIR (air.security) is a Tel Aviv-based startup founded in February 2026 by Yair Saban (CEO) and Niv Hoffman (CTO), both veterans of Israel’s Unit 8200 signals-intelligence unit. The company emerged from stealth on September 1, 2026 with $50 million raised across two seed rounds ($10M led by Sequoia Capital, followed by $40M led by Greenoaks), with participation from individual investors including Wiz co-founder Yinon Costica. Former Disney and Costco CISO Ryan Knisley joined as Chief Strategy Officer. At the time of its funding announcement the company reported roughly 40 employees and more than 20 paying customers, about a quarter of them large enterprises, concentrated in financial services and pharma.

The product, marketed as a ‘context firewall for AI agents,’ sits between enterprise AI agents and the external skills, plugins, and Model Context Protocol (MCP) servers those agents try to use. It has four components: Control (discovery and policy governance of agent fleets), Filter (pre-deployment vetting of add-ons for malicious instructions, excessive permissions, or supply-chain risk), Defend (runtime monitoring and blocking), and a curated Marketplace of pre-vetted add-ons. AIR’s own research, cited in press coverage, claims to have found over 17,800 public AI add-ons with 6.7 million installs relying on untrusted instruction sources, plus fake ‘skills’ impersonating Anthropic and OpenAI to slip past review.

The company’s differentiator is narrow focus: rather than general LLM security or prompt-injection filtering, it specifically targets the software-supply-chain risk created by autonomous agents installing third-party skills and MCP tools without human review. This is a genuine and fast-growing problem as agentic AI adoption accelerates, but AIR competes with several other very recently funded ‘AI agent security’ and ‘MCP security’ entrants, and as a six-month-old company it has no long-run track record.